In this notebook, a template is provided for you to implement your functionality in stages, which is required to successfully complete this project. If additional code is required that cannot be included in the notebook, be sure that the Python code is successfully imported and included in your submission if necessary.
Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.
In addition to implementing code, there is a writeup to complete. The writeup should be completed in a separate file, which can be either a markdown file or a pdf document. There is a write up template that can be used to guide the writing process. Completing the code template and writeup template will cover all of the rubric points for this project.
The rubric contains "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. The stand out suggestions are optional. If you decide to pursue the "stand out suggestions", you can include the code in this Ipython notebook and also discuss the results in the writeup file.
Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. In addition, Markdown cells can be edited by typically double-clicking the cell to enter edit mode.
import pickle
training_file = 'train.p'
validation_file = 'valid.p'
testing_file = 'test.p'
with open(training_file, mode='rb') as f:
train = pickle.load(f)
with open(validation_file, mode='rb') as f:
valid = pickle.load(f)
with open(testing_file, mode='rb') as f:
test = pickle.load(f)
X_train_org, y_train_org = train['features'], train['labels']
X_valid_org, y_valid_org = valid['features'], valid['labels']
X_test_org , y_test_org = test['features'], test['labels']
The pickled data is a dictionary with 4 key/value pairs:
'features' is a 4D array containing raw pixel data of the traffic sign images, (num examples, width, height, channels).'labels' is a 1D array containing the label/class id of the traffic sign. The file signnames.csv contains id -> name mappings for each id.'sizes' is a list containing tuples, (width, height) representing the original width and height the image.'coords' is a list containing tuples, (x1, y1, x2, y2) representing coordinates of a bounding box around the sign in the image. THESE COORDINATES ASSUME THE ORIGINAL IMAGE. THE PICKLED DATA CONTAINS RESIZED VERSIONS (32 by 32) OF THESE IMAGESComplete the basic data summary below. Use python, numpy and/or pandas methods to calculate the data summary rather than hard coding the results. For example, the pandas shape method might be useful for calculating some of the summary results.
### Replace each question mark with the appropriate value.
### Use python, pandas or numpy methods rather than hard coding the results
# Number of training examples
n_train = X_train_org.shape[0]
# Number of testing examples.
n_test = X_test_org.shape[0]
# What's the shape of an traffic sign image?
image_shape = X_train_org.shape[1:3]
# How many unique classes/labels there are in the dataset.
n_classes = max(y_train_org)+1
print("Number of training examples =", n_train)
print("Number of testing examples =", n_test)
print("Image data shape =", image_shape)
print("Number of classes =", n_classes)
Visualize the German Traffic Signs Dataset using the pickled file(s). This is open ended, suggestions include: plotting traffic sign images, plotting the count of each sign, etc.
The Matplotlib examples and gallery pages are a great resource for doing visualizations in Python.
NOTE: It's recommended you start with something simple first. If you wish to do more, come back to it after you've completed the rest of the sections.
### Data exploration visualization code goes here.
### Feel free to use as many code cells as needed.
import matplotlib.pyplot as plt
# Visualizations will be shown in the notebook.
%matplotlib inline
import pandas as pd
df = pd.read_csv('./signnames.csv')
df
def visualise_dataset(X_train,y_train):
for i in range(0,X_train.shape[0],int(X_train.shape[0]/400)):
image = X_train[i,:,:,:]
fig, ax = plt.subplots()
ax.imshow(image)
plt.show()
sign_id = y_train[i]
sign_name = df.values[int(sign_id),-1]
print('{1}, ID: {0}'.format(sign_id,sign_name))
visualise_dataset(X_train_org,y_train_org)
y_train_org[range(0,34799,400)]
Design and implement a deep learning model that learns to recognize traffic signs. Train and test your model on the German Traffic Sign Dataset.
The LeNet-5 implementation shown in the classroom at the end of the CNN lesson is a solid starting point. You'll have to change the number of classes and possibly the preprocessing, but aside from that it's plug and play!
With the LeNet-5 solution from the lecture, you should expect a validation set accuracy of about 0.89. To meet specifications, the validation set accuracy will need to be at least 0.93. It is possible to get an even higher accuracy, but 0.93 is the minimum for a successful project submission.
There are various aspects to consider when thinking about this problem:
Here is an example of a published baseline model on this problem. It's not required to be familiar with the approach used in the paper but, it's good practice to try to read papers like these.
Use the code cell (or multiple code cells, if necessary) to implement the first step of your project.
print(X_train_org.shape)
print(X_valid_org.shape)
print(X_test_org.shape)
print(X_train_org[0,0,0,:])
import cv2
import numpy as np
def one_hot_encode(x):
"""
One hot encode a list of sample labels. Return a one-hot encoded vector for each label.
: x: List of sample Labels
: return: Numpy array of one-hot encoded labels
"""
np.set_printoptions(threshold=np.nan)
from sklearn.preprocessing import LabelBinarizer
# Turn labels into numbers and apply One-Hot Encoding
encoder = LabelBinarizer()
encoder.fit(range(43))
x = encoder.transform(x)
# Change to float32, so it can be multiplied against the features in TensorFlow, which are float32
x = x.astype(np.float32)
return x
y_train = one_hot_encode(y_train_org)
y_valid = one_hot_encode(y_valid_org)
y_test = one_hot_encode(y_test_org)
def conv_to_concat_colorspace(image_batch):
shape = np.asarray(image_batch.shape)
shape[-1] = 4
concat_batch = np.empty(shape)
for idx in range(image_batch.shape[0]):
concat_batch[idx,:,:,0:3] = cv2.cvtColor(image_batch[idx], cv2.COLOR_RGB2YCrCb )
#concat_batch[idx,:,:,3:6] = cv2.cvtColor(image_batch[idx], cv2.COLOR_RGB2YUV )
#concat_batch[idx,:,:,3] = cv2.cvtColor(image_batch[idx], cv2.COLOR_RGB2HLS)[:,:,2]
concat_batch[idx,:,:,3] = cv2.cvtColor(image_batch[idx], cv2.COLOR_RGB2HSV)[:,:,1]
#concat_batch[idx,:,:,5:8] = image_batch[idx]
concat_batch[idx,:,:,:] = concat_batch[idx,:,:,:]/255
return concat_batch
use_YCrCB_S = False
if use_YCrCB_S:
%time X_train = conv_to_concat_colorspace(X_train_org)
%time X_valid = conv_to_concat_colorspace(X_valid_org)
%time X_test = conv_to_concat_colorspace(X_test_org)
else:
X_train = np.sum(X_train_org,axis=3, keepdims=True)/3/255
X_valid = np.sum(X_valid_org,axis=3, keepdims=True)/3/255
X_test = np.sum(X_test_org,axis=3, keepdims=True)/3/255
def visualise_dataset(image,y_valid_org,step_size):
plt_num = 1
print(image.shape)
for image_idx in range(0,image.shape[0],step_size):
channels = image.shape[3]
plt.figure(plt_num, figsize=(32,32))
for channel in range(channels):
plt.subplot(4,8, channel+1) # sets the number of feature maps to show on each row and column
plt.title('channel ' + str(channel)) # displays the feature map number
plt.imshow(image[image_idx,:,:, channel], interpolation="nearest", cmap="gray")
plt.show()
sign_id = y_train_org[image_idx]
sign_name = df.values[int(sign_id),-1]
print('{1}, ID: {0}'.format(sign_id,sign_name))
%time visualise_dataset(X_valid[:,:,:,:],y_valid_org,int(X_valid.shape[0]/15))
if False:
with open('X_train.pickle', 'rb') as handle:
X_train = pickle.load(handle, protocol=pickle.HIGHEST_PROTOCOL)
with open('X_valid.pickle', 'rb') as handle:
X_valid = pickle.load(handle, protocol=pickle.HIGHEST_PROTOCOL)
with open('X_test.pickle', 'rb') as handle:
X_test = pickle.load(handle, protocol=pickle.HIGHEST_PROTOCOL)
if False:
with open('X_train.pickle', 'wb') as handle:
pickle.dump(X_train, handle, protocol=pickle.HIGHEST_PROTOCOL)
with open('X_valid.pickle', 'wb') as handle:
pickle.dump(X_valid, handle, protocol=pickle.HIGHEST_PROTOCOL)
with open('X_test.pickle', 'wb') as handle:
pickle.dump(X_test, handle, protocol=pickle.HIGHEST_PROTOCOL)
print(X_train.shape)
print(X_valid.shape)
print(X_test.shape)
print(X_train[0,0,0,:])
import tensorflow as tf
def conv2d(x_tensor, conv_num_outputs, conv_ksize, conv_strides):
"""
Apply convolution then max pooling to x_tensor
:param x_tensor: TensorFlow Tensor
:param conv_num_outputs: Number of outputs for the convolutional layer
:param conv_ksize: kernal size 2-D Tuple for the convolutional layer
:param conv_strides: Stride 2-D Tuple for convolution
:param pool_ksize: kernal size 2-D Tuple for pool
:param pool_strides: Stride 2-D Tuple for pool
: return: A tensor that represents convolution and max pooling of x_tensor
"""
# Create the weight and bias
biases = tf.Variable(tf.zeros(conv_num_outputs))
# TF 1.0 :
#weights_depth = x_tensor.shape.as_list()[-1]
# TF 0.12 :
weights_depth = x_tensor.get_shape().as_list()[-1]
weights_dim = [conv_ksize[0], conv_ksize[1], weights_depth, conv_num_outputs]
weights = tf.Variable(tf.truncated_normal(weights_dim))
# Apply Convolution
conv_strides = [1, conv_strides[0], conv_strides[1], 1] # (batch, height, width, depth)
padding = 'VALID'
conv_layer = tf.nn.conv2d(x_tensor, weights, conv_strides, padding)
# Add bias
conv_layer = tf.nn.bias_add(conv_layer, biases)
# Apply activation function
conv_layer = tf.nn.relu(conv_layer)
return conv_layer
def pool(conv_layer, pool_ksize, pool_strides):
filter_shape = [1, pool_ksize[0], pool_ksize[1], 1]
pool_strides = [1, pool_strides[0], pool_strides[1], 1]
padding = 'VALID'
pool_max = tf.nn.max_pool(conv_layer, filter_shape, pool_strides, padding)
#pool_avg = tf.nn.avg_pool(conv_layer, filter_shape, pool_strides, padding)
#pool_max = tf.nn.fractional_max_pool(conv_layer, [1.0, 2.0, 2.0, 1.0])
#pool_avg = tf.nn.fractional_avg_pool(conv_layer, [1.0, 2.0, 2.0, 1.0])
#pool = tf.concat(3,[pool_max,pool_avg])
return pool_max
def flatten(x_tensor):
"""
Flatten x_tensor to (Batch Size, Flattened Image Size)
: x_tensor: A tensor of size (Batch Size, ...), where ... are the image dimensions.
: return: A tensor of size (Batch Size, Flattened Image Size).
"""
dimensions = (x_tensor.get_shape().as_list()[1:4])
prod = 1
for dimension in dimensions:
prod *= dimension
x_tensor = tf.reshape(x_tensor, [-1,prod])
return x_tensor
def fully_conn(x_tensor, num_outputs):
"""
Apply a fully connected layer to x_tensor using weight and bias
: x_tensor: A 2-D tensor where the first dimension is batch size.
: num_outputs: The number of output that the new tensor should be.
: return: A 2-D tensor where the second dimension is num_outputs.
"""
tensor_out = tf.contrib.layers.fully_connected(x_tensor, num_outputs)
return tensor_out
def conv_net(x, keep_prob):
"""
Create a convolutional neural network model
: x: Placeholder tensor that holds image data.
: keep_prob: Placeholder tensor that hold dropout keep probability.
: return: Tensor that represents logits
"""
x_tensor = x #:param x_tensor: TensorFlow Tensor
conv_strides = (1,1) #:param conv_strides: Stride 2-D Tuple for convolution
pool_ksize = (2,2) #:param pool_ksize: kernal size 2-D Tuple for pool
pool_strides = (1,1) #:param pool_strides: Stride 2-D Tuple for pool
conv_ksize = (3,3)
conv_num_outputs = 5 #:param conv_num_outputs: Number of outputs for the convolutional layer
x_tensor = conv2d(x_tensor, conv_num_outputs, conv_ksize, conv_strides)
conv_num_outputs = 7 #:param conv_num_outputs: Number of outputs for the convolutional layer
x_tensor = conv2d(x_tensor, conv_num_outputs, conv_ksize, conv_strides)
#x_tensor = pool(x_tensor, pool_ksize, pool_strides)
#conv_num_outputs = 17 #:param conv_num_outputs: Number of outputs for the convolutional layer
#x_tensor = conv2d(x_tensor, conv_num_outputs, conv_ksize, conv_strides)
#x_tensor = conv2d(x_tensor, conv_num_outputs, conv_ksize, conv_strides)
#x_tensor = pool(x_tensor, pool_ksize, pool_strides)
return x_tensor
def fully_con_net(x_tensor, keep_prob):
x_tensor = flatten(x_tensor)
num_outputs = n_classes*4
x_tensor = fully_conn(x_tensor, int(num_outputs))
x_tensor = tf.nn.dropout(x_tensor, keep_prob)
num_outputs = n_classes*3
x_tensor = fully_conn(x_tensor, int(num_outputs))
x_tensor = tf.nn.dropout(x_tensor, keep_prob)
num_outputs = n_classes*2
x_tensor = tf.nn.dropout(x_tensor, keep_prob)
x_tensor = fully_conn(x_tensor, int(num_outputs))
num_outputs = n_classes
x_tensor = fully_conn(x_tensor, int(num_outputs))
return x_tensor
def vgg_like(x_pre):
# conv + pool
conv_ksize = (3,3)
conv_strides = (1,1) #:param conv_strides: Stride 2-D Tuple for convolution
pool_ksize = (2,2) #:param pool_ksize: kernal size 2-D Tuple for pool
pool_strides = (1,1) #:param pool_strides: Stride 2-D Tuple for pool
conv_num_outputs = 9 #:param conv_num_outputs: Number of outputs for the convolutional layer
c00 = conv2d(x_pre, conv_num_outputs, conv_ksize, conv_strides)
conv_num_outputs = 9 #:param conv_num_outputs: Number of outputs for the convolutional layer
c01 = conv2d(c00, conv_num_outputs, conv_ksize, conv_strides)
p0 = pool(c01, pool_ksize, pool_strides)
conv_num_outputs = 9 #:param conv_num_outputs: Number of outputs for the convolutional layer
c10 = conv2d(x_pre, conv_num_outputs, conv_ksize, conv_strides)
conv_num_outputs = 9 #:param conv_num_outputs: Number of outputs for the convolutional layer
c11 = conv2d(c00, conv_num_outputs, conv_ksize, conv_strides)
p1 = pool(c01, pool_ksize, pool_strides)
tf_activation = p1
# pick layers
p1f = flatten(p1)
p0f = flatten(p0)
concat = tf.concat(1,[p1f,p0f])
# fully connected layers
num_outputs = n_classes*4
x_tensor = fully_conn(concat, int(num_outputs))
x_tensor = tf.nn.dropout(x_tensor, keep_prob)
num_outputs = n_classes
x_tensor = fully_conn(x_tensor, int(num_outputs))
logits = fully_con_net(x_tensor, keep_prob)
return logits
def LeNet(x):
# Arguments used for tf.truncated_normal, randomly defines variables for the weights and biases for each layer
mu = 0
sigma = 0.1
# SOLUTION: Layer 1: Convolutional. Input = 32x32x1. Output = 28x28x6.
conv1_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 1, 6), mean = mu, stddev = sigma))
conv1_b = tf.Variable(tf.zeros(6))
conv1 = tf.nn.conv2d(x, conv1_W, strides=[1, 1, 1, 1], padding='VALID') + conv1_b
# SOLUTION: Activation.
conv1 = tf.nn.relu(conv1)
# SOLUTION: Pooling. Input = 28x28x6. Output = 14x14x6.
conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
# SOLUTION: Layer 2: Convolutional. Output = 10x10x16.
conv2_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 6, 16), mean = mu, stddev = sigma))
conv2_b = tf.Variable(tf.zeros(16))
conv2 = tf.nn.conv2d(conv1, conv2_W, strides=[1, 1, 1, 1], padding='VALID') + conv2_b
# SOLUTION: Activation.
conv2 = tf.nn.relu(conv2)
# SOLUTION: Pooling. Input = 10x10x16. Output = 5x5x16.
conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
tf_activation = conv2
# SOLUTION: Flatten. Input = 5x5x16. Output = 400.
fc0 = flatten(conv2)
# SOLUTION: Layer 3: Fully Connected. Input = 400. Output = 120.
fc1_W = tf.Variable(tf.truncated_normal(shape=(400, 120), mean = mu, stddev = sigma))
fc1_b = tf.Variable(tf.zeros(120))
fc1 = tf.matmul(fc0, fc1_W) + fc1_b
# SOLUTION: Activation.
fc1 = tf.nn.relu(fc1)
# SOLUTION: Layer 4: Fully Connected. Input = 120. Output = 84.
fc2_W = tf.Variable(tf.truncated_normal(shape=(120, 84), mean = mu, stddev = sigma))
fc2_b = tf.Variable(tf.zeros(84))
fc2 = tf.matmul(fc1, fc2_W) + fc2_b
# SOLUTION: Activation.
fc2 = tf.nn.relu(fc2)
# SOLUTION: Layer 5: Fully Connected. Input = 84. Output = 10.
fc3_W = tf.Variable(tf.truncated_normal(shape=(84, n_classes), mean = mu, stddev = sigma))
fc3_b = tf.Variable(tf.zeros(n_classes))
logits = tf.matmul(fc2, fc3_W) + fc3_b
return logits, tf_activation
def LeNet42(x):
# Arguments used for tf.truncated_normal, randomly defines variables for the weights and biases for each layer
mu = 0
sigma = 1/16
x = tf.nn.local_response_normalization(x)
# Layer 1: Convolutional. Input = 32x32x1. Output = 28x28x6.
depth_1 = 6
conv1_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 1, depth_1), mean = mu, stddev = sigma))
conv1_b = tf.Variable(tf.zeros(depth_1))
conv1 = tf.nn.conv2d(x, conv1_W, strides=[1, 1, 1, 1], padding='VALID') + conv1_b
# Activation.
conv1 = tf.nn.relu(conv1)
# Pooling. Input = 28x28x6. Output = 14x14x6.
conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
# Layer 2: Convolutional. Output = 10x10x16.
depth_2 = 42
conv2_W = tf.Variable(tf.truncated_normal(shape=(5, 5, depth_1, depth_2), mean = mu, stddev = sigma))
conv2_b = tf.Variable(tf.zeros(depth_2))
conv2 = tf.nn.conv2d(conv1, conv2_W, strides=[1, 1, 1, 1], padding='VALID') + conv2_b
# Activation.
conv2 = tf.nn.relu(conv2)
# Pooling. Input = 10x10x16. Output = 5x5x16.
conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
tf_activation = conv2
# Flatten. Input = 5x5x16. Output = 400.
fc0 = flatten(conv2)
# Layer 3: Fully Connected.
depth_in = int(fc0.get_shape()[1])
depth_out = int(depth_in * 120/400)
fc1_W = tf.Variable(tf.truncated_normal(shape=(depth_in, depth_out), mean = mu, stddev = sigma))
fc1_b = tf.Variable(tf.zeros(depth_out))
fc1 = tf.matmul(fc0, fc1_W) + fc1_b
# Activation.
fc1 = tf.nn.relu(fc1)
# Layer 4: Fully Connected.
depth_in = int(fc1.get_shape()[1])
depth_out = int(depth_in * 84/120)
fc2_W = tf.Variable(tf.truncated_normal(shape=(depth_in, depth_out), mean = mu, stddev = sigma))
fc2_b = tf.Variable(tf.zeros(depth_out))
fc2 = tf.matmul(fc1, fc2_W) + fc2_b
fc2 = tf.nn.dropout(fc2, keep_prob)
# Activation.
fc2 = tf.nn.relu(fc2)
# Layer 5: Fully Connected.
depth_in = int(fc2.get_shape()[1])
depth_out = n_classes
fc3_W = tf.Variable(tf.truncated_normal(shape=(depth_in, depth_out), mean = mu, stddev = sigma))
fc3_b = tf.Variable(tf.zeros(depth_out))
fc3 = tf.matmul(fc2, fc3_W) + fc3_b
# Activation.
#fc3 = tf.nn.relu(fc3)
# Layer 6: Fully Connected.
#depth_in = int(fc3.get_shape()[1])
#fc4_W = tf.Variable(tf.truncated_normal(shape=(depth_in, n_classes), mean = mu, stddev = sigma))
#fc4_b = tf.Variable(tf.zeros(n_classes))
#logits = tf.matmul(fc3, fc4_W) + fc4_b
logits = fc3
return logits, tf_activation
##############################
## Build the Neural Network ##
##############################
# Remove previous weights, bias, inputs, etc..
tf.reset_default_graph()
# Inputs
image_shape = X_train.shape[1:]
x = tf.placeholder(tf.float32, shape=[None, image_shape[0], image_shape[1], image_shape[2]],name = 'x')
y = tf.placeholder(tf.float32, shape=[None, n_classes],name = 'y')
keep_prob = tf.placeholder(tf.float32, name='keep_prob')
# Model
x_pre = x#tf.nn.dropout(x, keep_prob) #:param x_tensor: TensorFlow Tensor
x_pre = tf.nn.local_response_normalization(x_pre)
logits, tf_activation = LeNet42(x_pre)
# Name logits Tensor, so that is can be loaded from disk after training
logits = tf.identity(logits, name='logits')
# Loss and Optimizer
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=y))
optimizer = tf.train.AdamOptimizer().minimize(cost)
#optimizer = tf.train.RMSPropOptimizer(0.001).minimize(cost)
# Accuracy
correct_pred = tf.equal(tf.argmax(logits, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32), name='accuracy')
A validation set can be used to assess how well the model is performing. A low accuracy on the training and validation sets imply underfitting. A high accuracy on the training set but low accuracy on the validation set implies overfitting.
### Train your model here.
### Calculate and report the accuracy on the training and validation set.
### Once a final model architecture is selected,
### the accuracy on the test set should be calculated and reported as well.
### Feel free to use as many code cells as needed.
def train_neural_network(session, optimizer, keep_probability, feature_batch, label_batch):
"""
Optimize the session on a batch of images and labels
: session: Current TensorFlow session
: optimizer: TensorFlow optimizer function
: keep_probability: keep probability
: feature_batch: Batch of Numpy image data
: label_batch: Batch of Numpy label data
"""
#x = neural_net_image_input((32, 32, 3))
#y = neural_net_label_input(10)
#keep_prob = neural_net_keep_prob_input()
#print(label_batch)
config=tf.ConfigProto(#allow_soft_placement=True,
#log_device_placement=True,
# device_count = {'GPU': 8}
)
sess = tf.Session(config=config)
feed_dict={keep_prob:keep_probability,x:feature_batch,y:label_batch}
#print(feed_dict)
session.run(optimizer,feed_dict=feed_dict)
def print_stats(session, feature_batch, label_batch, cost, accuracy):
"""
Print information about loss and validation accuracy
: session: Current TensorFlow session
: feature_batch: Batch of Numpy image data
: label_batch: Batch of Numpy label data
: cost: TensorFlow cost function
: accuracy: TensorFlow accuracy function
"""
feed_dict={keep_prob:1.,x:feature_batch,y:label_batch}
# Should this be done on validation data?
# Calculate batch loss and accuracy
loss = session.run(cost,feed_dict=feed_dict)
valid_acc = sess.run(accuracy,feed_dict=feed_dict)
#print('Loss: {0:9.5f} Accuracy: {1:9.5f}'.format(loss,valid_acc))
return loss, valid_acc
def get_next_batch(batches, batch_size = 64):
n_batches = len(batches)//batch_size
for idx in range(0,n_batches,batch_size):
batch = batches[idx:idx+batch_size,:,:,:]
yield [x,y]
# Parameters
epochs = 100 # 1000 @ 128 => 3430 sec / ~1h
batch_size = 1000
keep_probability = 0.25
saved_acc = .0
save_model_path = './'
saver = tf.train.Saver()
import time
start = time.time()
print('Training...')
with tf.Session() as sess:
# Initializing the variables
try:
saver.restore(sess, save_model_path)
print('Restored Model.')
except:
sess.run(tf.global_variables_initializer())
print('Initialised Model...')
# Training cycle
for epoch in range(epochs):
# Loop over all batches
for idx in range(0,int(len(X_train)/batch_size)):
n_batches = len(X_train)//batch_size
batch_idx = ((np.random.random(batch_size)*X_train.shape[0])).astype(int)
batch_features = X_train[batch_idx,:,:,:]
batch_labels = y_train[batch_idx]
train_neural_network(sess, optimizer, keep_probability, batch_features, batch_labels)
if idx % 5 == 0:
loss, acc = print_stats(sess, batch_features, batch_labels, cost, accuracy)
loss_valid, acc_valid = print_stats(sess, X_valid, y_valid, cost, accuracy)
print('Epoch {:>4}, Batch {:>6}: Loss: {:9.5f} Accuracy: {:9.5f} Validation: Loss: {:9.5f} Accuracy: {:9.5f}'.format(
epoch + 1, idx,loss,acc,loss_valid, acc_valid))
if acc_valid > saved_acc:
try:
save_path = saver.save(sess, save_model_path)
saved_acc = acc_valid
print('Model saved')
except:
print('Model saving failed')
if acc_valid > 0.93:
break
print('Training time : {}'.format(time.time() - start))
To give yourself more insight into how your model is working, download at least five pictures of German traffic signs from the web and use your model to predict the traffic sign type.
You may find signnames.csv useful as it contains mappings from the class id (integer) to the actual sign name.
### Load the images and plot them here.
### Feel free to use as many code cells as needed.
visualise_dataset(X_test[:,:,:,:],y_test)
save_model_path = './'
n_samples = 12
top_n_predictions = 5
import random
import pandas as pd
from sklearn.preprocessing import LabelBinarizer
def _load_label_names():
"""
Load the label names from file
"""
df = pd.read_csv('./signnames.csv')
return df['SignName'].values
def display_image_predictions(features, labels, predictions,n_samples,top_n_predictions):
n_classes = 43
label_names = _load_label_names()
label_binarizer = LabelBinarizer()
label_binarizer.fit(range(n_classes))
label_ids = label_binarizer.inverse_transform(np.array(labels))
fig, axies = plt.subplots(nrows=n_samples, ncols=2)
#fig.tight_layout()
#fig.suptitle('Softmax Predictions', fontsize=20, y=1.1)
fig.suptitle('Softmax Predictions', fontsize=44)
fig.set_size_inches(22, n_samples*2)
n_predictions = top_n_predictions ### Check this later, computer is busy
margin = 0.05
ind = np.arange(n_predictions)
width = (1. - 2. * margin) / n_predictions
for image_i, (feature, label_id, pred_indicies, pred_values) in enumerate(zip(
features, label_ids, predictions.indices, predictions.values)):
pred_names = [label_names[pred_i] for pred_i in pred_indicies]
correct_name = label_names[label_id]
axies[image_i][0].imshow(feature[:,:,0], interpolation="nearest", cmap="gray")
axies[image_i][0].set_title(correct_name)
axies[image_i][0].set_axis_off()
axies[image_i][1].barh(ind + margin, pred_values[::-1], width)
axies[image_i][1].set_yticks(ind + margin)
axies[image_i][1].set_yticklabels(pred_names[::-1])
axies[image_i][1].set_xticks([0, 0.5, 1.0])
def test_model(X_test,y_test,batch_size,random_sampels=True):
"""
Test the saved model against the test dataset
"""
test_features, test_labels = X_test, y_test
loaded_graph = tf.Graph()
with tf.Session(graph=loaded_graph) as sess:
# Load model
loader = tf.train.import_meta_graph(save_model_path + '.meta')
loader.restore(sess, save_model_path)
# Get Tensors from loaded model
loaded_x = loaded_graph.get_tensor_by_name('x:0')
loaded_y = loaded_graph.get_tensor_by_name('y:0')
loaded_keep_prob = loaded_graph.get_tensor_by_name('keep_prob:0')
loaded_logits = loaded_graph.get_tensor_by_name('logits:0')
loaded_acc = loaded_graph.get_tensor_by_name('accuracy:0')
# Get accuracy in batches for memory limitations
test_batch_acc_total = 0
test_batch_count = 0
#for train_feature_batch, train_label_batch in helper.batch_features_labels(test_features, test_labels, batch_size):
n_batches = len(X_test)//batch_size
for idx in range(0,n_batches,batch_size):
train_feature_batch = X_test[idx:idx+batch_size,:,:,:]
train_label_batch = y_test[idx:idx+batch_size]
test_batch_acc_total += sess.run(
loaded_acc,
feed_dict={loaded_x: train_feature_batch, loaded_y: train_label_batch, loaded_keep_prob: 1.0})
test_batch_count += 1
print('Testing Accuracy: {}\n'.format(test_batch_acc_total/test_batch_count))
# Print Random Samples
if random_sampels :
random_test_features, random_test_labels = tuple(zip(*random.sample(list(zip(test_features, test_labels)), n_samples)))
else:
random_test_features, random_test_labels = X_test,y_test
random_test_predictions = sess.run(
tf.nn.top_k(tf.nn.softmax(loaded_logits), top_n_predictions),
feed_dict={loaded_x: random_test_features, loaded_y: random_test_labels, loaded_keep_prob: 1.0})
display_image_predictions(random_test_features, random_test_labels, random_test_predictions,n_samples,top_n_predictions)
return random_test_predictions
_ = test_model(X_test,y_test,batch_size)
### Calculate the accuracy for these 5 new images.
### For example, if the model predicted 1 out of 5 signs correctly, it's 20% accurate on these new images.
n_samples = 5
webimages = np.empty([n_samples,32,32,3])
webimages[0] = cv2.imread("webimages/1.jpg")
webimages[1] = cv2.imread("webimages/2.jpg")
webimages[2] = cv2.imread("webimages/3.jpg")
webimages[3] = cv2.imread("webimages/4.jpg")
webimages[4] = cv2.imread("webimages/5.jpg")
#visualise_dataset(webimages[:,:,:,:],y_test,1)
np.sum(webimages,axis=3, keepdims=True)/3/255
random_test_predictions = test_model(webimages_gray,one_hot_encode([23,17,14,26,1]),batch_size=5,random_sampels=False)
#print(random_test_predictions.values)
For each of the new images, print out the model's softmax probabilities to show the certainty of the model's predictions (limit the output to the top 5 probabilities for each image). tf.nn.top_k could prove helpful here.
The example below demonstrates how tf.nn.top_k can be used to find the top k predictions for each image.
tf.nn.top_k will return the values and indices (class ids) of the top k predictions. So if k=3, for each sign, it'll return the 3 largest probabilities (out of a possible 43) and the correspoding class ids.
Take this numpy array as an example. The values in the array represent predictions. The array contains softmax probabilities for five candidate images with six possible classes. tk.nn.top_k is used to choose the three classes with the highest probability:
# (5, 6) array
a = np.array([[ 0.24879643, 0.07032244, 0.12641572, 0.34763842, 0.07893497,
0.12789202],
[ 0.28086119, 0.27569815, 0.08594638, 0.0178669 , 0.18063401,
0.15899337],
[ 0.26076848, 0.23664738, 0.08020603, 0.07001922, 0.1134371 ,
0.23892179],
[ 0.11943333, 0.29198961, 0.02605103, 0.26234032, 0.1351348 ,
0.16505091],
[ 0.09561176, 0.34396535, 0.0643941 , 0.16240774, 0.24206137,
0.09155967]])
Running it through sess.run(tf.nn.top_k(tf.constant(a), k=3)) produces:
TopKV2(values=array([[ 0.34763842, 0.24879643, 0.12789202],
[ 0.28086119, 0.27569815, 0.18063401],
[ 0.26076848, 0.23892179, 0.23664738],
[ 0.29198961, 0.26234032, 0.16505091],
[ 0.34396535, 0.24206137, 0.16240774]]), indices=array([[3, 0, 5],
[0, 1, 4],
[0, 5, 1],
[1, 3, 5],
[1, 4, 3]], dtype=int32))
Looking just at the first row we get [ 0.34763842, 0.24879643, 0.12789202], you can confirm these are the 3 largest probabilities in a. You'll also notice [3, 0, 5] are the corresponding indices.
### Print out the top five softmax probabilities for the predictions on the German traffic sign images found on the web.
### Feel free to use as many code cells as needed.
This Section is not required to complete but acts as an additional excersise for understaning the output of a neural network's weights. While neural networks can be a great learning device they are often referred to as a black box. We can understand what the weights of a neural network look like better by plotting their feature maps. After successfully training your neural network you can see what it's feature maps look like by plotting the output of the network's weight layers in response to a test stimuli image. From these plotted feature maps, it's possible to see what characteristics of an image the network finds interesting. For a sign, maybe the inner network feature maps react with high activation to the sign's boundary outline or to the contrast in the sign's painted symbol.
Provided for you below is the function code that allows you to get the visualization output of any tensorflow weight layer you want. The inputs to the function should be a stimuli image, one used during training or a new one you provided, and then the tensorflow variable name that represents the layer's state during the training process, for instance if you wanted to see what the LeNet lab's feature maps looked like for it's second convolutional layer you could enter conv2 as the tf_activation variable.
For an example of what feature map outputs look like, check out NVIDIA's results in their paper End-to-End Deep Learning for Self-Driving Cars in the section Visualization of internal CNN State. NVIDIA was able to show that their network's inner weights had high activations to road boundary lines by comparing feature maps from an image with a clear path to one without. Try experimenting with a similar test to show that your trained network's weights are looking for interesting features, whether it's looking at differences in feature maps from images with or without a sign, or even what feature maps look like in a trained network vs a completely untrained one on the same sign image.
Your output should look something like this (above)
### Visualize your network's feature maps here.
### Feel free to use as many code cells as needed.
# image_input: the test image being fed into the network to produce the feature maps
# tf_activation: should be a tf variable name used during your training procedure that represents the calculated state of a specific weight layer
# activation_min/max: can be used to view the activation contrast in more detail, by default matplot sets min and max to the actual min and max values of the output
# plt_num: used to plot out multiple different weight feature map sets on the same block, just extend the plt number for each new feature map entry
def outputFeatureMap(image_input, tf_activation, activation_min=-1, activation_max=-1 ,plt_num=1):
# Here make sure to preprocess your image_input in a way your network expects
# with size, normalization, ect if needed
# image_input =
# Note: x should be the same name as your network's tensorflow data placeholder variable
# If you get an error tf_activation is not defined it maybe having trouble accessing the variable from inside a function
activation = tf_activation.eval(session=sess,feed_dict={x : image_input,keep_prob:1})
print(activation.shape)
featuremaps = activation.shape[3]
plt.figure(plt_num, figsize=(15,15))
for featuremap in range(featuremaps):
plt.subplot(6,8, featuremap+1) # sets the number of feature maps to show on each row and column
plt.title('FeatureMap ' + str(featuremap)) # displays the feature map number
if activation_min != -1 & activation_max != -1:
plt.imshow(activation[0,:,:, featuremap], interpolation="nearest", vmin =activation_min, vmax=activation_max, cmap="gray")
elif activation_max != -1:
plt.imshow(activation[0,:,:, featuremap], interpolation="nearest", vmax=activation_max, cmap="gray")
elif activation_min !=-1:
plt.imshow(activation[0,:,:, featuremap], interpolation="nearest", vmin=activation_min, cmap="gray")
else:
plt.imshow(activation[0,:,:, featuremap], interpolation="nearest", cmap="gray")
image = X_train[-1:,:,:,:]
fig, ax = plt.subplots()
ax.imshow(image[0,:,:,0],interpolation="nearest", cmap="gray")
plt.show()
with tf.Session() as sess:
# Initializing the variables
try:
saver.restore(sess, save_model_path)
print('Restored Model...')
except:
sess.run(tf.global_variables_initializer())
print('Initialised Model...')
outputFeatureMap(image, tf_activation, activation_min=-1, activation_max=-1 ,plt_num=1)
Discuss how you used the visual output of your trained network's feature maps to show that it had learned to look for interesting characteristics in traffic sign images
Answer:
Note: Once you have completed all of the code implementations and successfully answered each question above, you may finalize your work by exporting the iPython Notebook as an HTML document. You can do this by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.
##Writeup Template
Build a Traffic Sign Recognition Project
The goals / steps of this project are the following:
The code for this step is contained in the second code cell of the IPython notebook.
I used the pandas library to load the signnames
Number of training examples = 34799 Number of testing examples = 12630 Image data shape = (32, 32) Number of classes = 43
I just plottet the images, ids and sign names. I know this signs very well so i could also verify that the labels are correct.
The model did learn very fast but stuck early
then i tried a VGG like arhitecture als so with not good enough results
after that i modifiyed the LeNet Architecture
going wirder did work
after still not getting good enough results i added normalisiotion to the model input and dropout in the dense layers and retrained the model with succsess
i did go with the existing data splits.
There was a lot of data, it seems it comes from videosequences moving mostly towards the signs.
I jused random batches to get rid of the sorting/sequencing.
Because of the bigger model i reduced the std_dev for the inits to 1/16.
The code for my final model is located in the seventh cell of the ipython notebook.
x = tf.nn.local_response_normalization(x)
My final model consisted of the following layers:
| Layer | Description |
|---|---|
| Input | 32x32x1 Gray image |
| Normalisation | i used local_response_normalization |
| Convolution 3x3 | 1x1 stride, valid padding, outputs 28x28x6 |
| RELU | |
| Max pooling | 2x2 stride, outputs 14x14x6 |
| Convolution 3x3 | 1x1 stride, valid padding, outputs 14x14x42. |
| RELU | |
| Max pooling | 2x2 stride, outputs 7x7x42 |
| Fully connected x3 Layers | i calculated the width depending on the input width. |
| Softmax | |
My Hyperparameters gets tuned all the time i end up with:
epochs = 100
batch_size = 1000
keep_probability = 0.25
the epochs loop exits when the needed accuracy was hit
I tried the RMSPropOptimizer but end up with AdamOptimizer
The code for calculating the accuracy of the model is located in the ninth cell of the Ipython notebook.
My final model results were:
I tryed a lot of diffrent approches Some didn train at all, others did not reach enough accuracy Best results i get by adopting the width of the layers my maintaining the proportions of LeNet To avoid overfitting i retrained with normalisation at the inputlayer and increasing dropout
Here are five German traffic signs that I found on the web:

All pictures had some difficultys 1: just signs in the background 2: overpainted 3: other text 4: small, other signs on picture, easy to confuse in grayscale 5: rotated, distorted, broken, on ground, dirty
The code for making predictions on my final model is located in the tenth cell of the Ipython notebook.
Here are the results of the prediction:
| Image | Prediction |
|---|---|
| slippery road | slippery road |
| no entry | no entry |
| stop | stop |
| traffic signals | general caution |
| speedlimit 30 | speedlimit 50 |
The model was able to correctly guess 2 of the 5 traffic signs, which gives an accuracy of 60%. This compares favorably to the accuracy on the test set of ...
The code for making predictions on my final model is located in the 11th cell of the Ipython notebook.
For the first image, the model is relatively sure that this is a stop sign (probability of 0.6), and the image does contain a stop sign. The top five soft max probabilities were
| Probability | Prediction |
|---|---|
| 0.977029383 | slippery road |
| 0.996126950 | no entry |
| 0.702272236 | stop |
| 0.447650850 | general caution |
| 0.797175407 | speedlimit 50 |
compleate np array:
[[ 9.77029383e-01 2.19183285e-02 8.18954257e-04 1.74098066e-04 5.30776197e-05] [ 9.96126950e-01 3.33242235e-03 4.47420840e-04 8.90796800e-05 1.77997424e-06] [ 7.02272236e-01 1.48365945e-01 9.19106230e-02 2.69952789e-02 1.24656009e-02] [ 4.47650850e-01 4.44780976e-01 8.56028497e-02 1.49302557e-02 5.40654548e-03] [ 7.97175407e-01 1.86552942e-01 5.91719337e-03 3.18665686e-03 2.19294964e-03]]